Two traders buy the same breakout on the same day. One is long AMD at 220. The other is long NVDA at 183. Both set a 3 percent stop, because a friend once told them 3 percent was a sensible number. AMD moves six dollars by lunch on normal chop and stops them out at 213.40. NVDA drifts two dollars sideways and nothing happens. The AMD trader blames the market. The AMD trader should blame the stop. A fixed percent rule pretends every stock breathes the same way. They do not. ATR bands are the fix, and a 1.5x to 2x ATR band is the rule I use for almost every swing trade I take.
This guide walks through ATR stop placement the way I actually run it on a live account. Where to anchor the band, why 1.5x to 2x is a range rather than a magic number, how the stop should move once the trade is working, and where the rule breaks. Examples are from real names this year, not clean textbook fantasies.
What an ATR Band Actually Measures
ATR is the Average True Range. It is the average size of a stock’s daily price move over a lookback window, corrected for gaps. Wilder’s original recipe uses a 14-day smoothed average, and that is still the setting I use. On a 14-day Wilder ATR, each day’s true range gets roughly 1/14 of the weight and the prior smoothed value carries the rest.
The formal definition of true range is the largest of these three on any given day:
TR = \max(H - L,\ |H - C_{prev}|,\ |L - C_{prev}|)Then ATR is the Wilder smoothing of TR over 14 periods:
ATR_t = ATR_{t-1} + \frac{1}{14}(TR_t - ATR_{t-1})An ATR band is simply the current ATR projected above or below a price anchor. Chandelier stops anchor to the highest high of the last N bars. Keltner channels anchor to an EMA. My swing stops anchor to the entry price on day one and to the highest close since entry on every day after. The distance is ATR times a multiplier, and the multiplier is where the work is.
The point of all this is one sentence. One dollar of AMD is not one dollar of NVDA. A 3 percent stop on a high-volatility name is noise. A 3 percent stop on a quiet name is a ransom note. An ATR band normalizes the stop to the stock’s own daily breathing, which is the only baseline that matters for your trade.
Why 1.5x to 2x Is a Range, Not a Magic Number
The range exists because two different things pull the multiplier in opposite directions. A tighter band gives you a better reward-to-risk ratio on paper, because the same target implies less risk per share. A looser band gives the stock room to shake and wiggle without taking you out on nothing. You cannot have both. You have to pick where on the spectrum a given trade sits.
My defaults look like this. On a clean breakout from a tight multi-week base, I use 1.5x ATR. The structure is telling me buyers have taken control, and I want the stop close enough that a failure is cheap to admit. On a later-stage entry, a pullback inside an established trend, or a move in a news-sensitive name, I widen to 2.0x ATR. The stock is either already extended or more likely to whip intraday on headlines, and I would rather eat a larger but rarer loss than get machine-gunned out of a real trend.
I rarely go below 1.5x. At 1.0x you are buying the right to be stopped out by one ordinary day. The math does not leave room for any mean reversion at all, and one 50-cent tape bomb during lunch takes the trade off your screen. On the upper side, I rarely go above 2.5x. Past that the stop is so wide that your position size has to collapse to keep risk constant, which means a winning trade barely moves the needle on your account even when you nail it.
If I cannot decide between 1.5 and 2, I use 1.75x. That is not a clever rule. It is just what a forty year old trader does when two reasonable defaults are both defensible and the clock is ticking.
Anchoring the Band to the Right Price
The multiplier is half the job. The anchor is the other half. A 2x ATR band dropped under the wrong price point is still a bad stop. My rules for the anchor are deliberately boring.
On entry day, anchor the band to the entry price. If I buy AMD at 220.27 with an ATR of 8.92, my initial stop is 220.27 minus 1.5 times 8.92, which is 206.89. Not below the base. Not below a swing low picked with a crayon. Below my actual cost, by a volatility-matched distance. The reason is simple. If the stock trades through my ATR band, I was wrong about the timing, and nothing about the base low tells me otherwise.
On every day after entry, move the anchor to the highest close since entry, and recompute the band using the latest ATR. If NVDA closes at 188.63 on a day when ATR is 5.38, the trailing stop is 188.63 minus 1.5 times 5.38, which is 180.56. If the next close is higher, the anchor moves up. If the next close is lower than the prior high-water mark, the anchor stays put. The stop only ratchets one way. This gives the trade room to pull back a full volatility unit without punishing you for a single red candle.
Anchoring to the daily close, not the intraday high, is deliberate. Intraday spikes are rarely settled price. I do not want a false wick to set my trailing anchor six dollars above where the stock actually wants to trade. The close is the market’s voted price for the day. Trust it.
Worked Example: AMD March 25 Breakout
AMD broke out of a three-week base on March 25, 2026. The prior range high was 209.11 on March 23. On March 25, AMD opened at 211.51, pushed to a high of 221.33, and closed at 220.27 on volume of 48.5 million shares, which cleared the 20-day volume baseline and qualified as a valid breakout on confirmed volume. A trader watching the tape would enter at or near the close.
The Wilder 14-day ATR ending March 24 was 8.92. Three stop choices fell out of that:
At 1.5x ATR, the stop sat at 220.27 minus 13.38, which is 206.89. That was 6.1 percent from entry and sat cleanly below the 209 breakout line, meaning the stop would survive a retest of the level but not a real failure.
At 2.0x ATR, the stop sat at 220.27 minus 17.84, which is 202.43. That was 8.1 percent from entry and sat well below the prior range, giving the trade room to churn after the initial thrust.
I would have used 1.5x on this entry. A breakout from a tight base with confirmed volume is exactly the situation where I want a tight volatility stop, because if it fails I want to know fast and the bull case is already in doubt.
The next session, March 26, AMD opened at 217.98, ran briefly to 221, and closed at 203.77, back below the 209 breakout level. The 1.5x ATR stop at 206.89 was broken on the close. A trader following the rule is flat the next morning, down roughly 6 percent on the share line, and has preserved the capital to try again. A trader using 2.0x ATR at 202.43 is technically still in, but the trade is already upside down and the close below the breakout level is a second signal. I treat that pair as an exit regardless.
The fake-out is not a knock on the rule. The point of the ATR band is not to guarantee you do not lose on a failed breakout. The point is that when you do lose, the loss is sized to the stock’s own noise floor, not to a generic percent that some newsletter recommended.
Worked Example: TSLA Trend Day on April 10
TSLA is the opposite of AMD in one important respect. It moves. On April 10, 2026, TSLA closed at 348.95 with a Wilder 14-day ATR of 14.80. A 1.5x stop is 22.20 wide. A 2.0x stop is 29.61 wide. Both are huge in absolute dollars compared to AMD, and both have to be huge, because TSLA routinely prints single-day ranges of 15 to 25 dollars on news that has no fundamental significance.
Imagine entering long at 348.95 on a pullback inside an established uptrend. That is a 2.0x ATR setup in my book, because TSLA whips hard on any tweet, earnings leak, or Elon rumor, and a 1.5x band often gets swept in normal trade. The initial stop sits at 348.95 minus 29.61, which is 319.34. That is 8.5 percent below entry.
Compare the two trades side by side. A 1.5x AMD stop risks 6.1 percent. A 2.0x TSLA stop risks 8.5 percent. Both are measured in the same language, a multiple of each stock’s own daily breathing. Neither trade is riskier than the other in volatility-adjusted terms. If I use a flat 1 percent of account risk on both, the AMD position is larger in share count and the TSLA position is smaller, and that is exactly the right outcome. The market does not know what either stock is worth in dollars. It only knows how hard each one swings.
On a quieter name like SPY, the same math gives very different dollar numbers. SPY closed April 10 at 679.46 with a Wilder ATR of 9.94. A 1.5x stop is 14.91 wide, or about 2.2 percent. On the same day, TSLA’s 2.0x stop was 8.5 percent. If a trader used a fixed 3 percent stop across all three, it would be three times too wide on SPY and far too tight on TSLA. That is not a small error. It is the entire reason ATR-based stops exist.
Trailing the Band Day by Day
Once a trade is working, I ratchet the ATR band forward each session using two inputs. The new anchor is the highest daily close since entry. The new distance is 1.5 or 2.0 times the latest ATR, using whichever multiplier I opened the trade with. The stop only moves up, never down.
This matters when volatility expands. On a news day, ATR can jump by 20 percent in a single session. If I let the stop widen in proportion, I give back open profit for no reason. The anti-widening rule is that the trailing stop can never move backwards. If yesterday’s computed stop was higher than today’s, yesterday’s stays in force. Some charting packages handle this automatically with a chandelier exit function. Others do not. Check your tool.
A related question is whether to trail on the close or on an intraday stop order. I trail on the close. End of day, I compute the new band, I put a resting stop at that level in my platform for tomorrow, and I do not touch it during the session. Intraday, the stock can do whatever it likes. I only care where it settles. This is not a rule I invented. It is a concession to the fact that daytime me makes worse decisions than evening me, and I protect the evening version by taking daytime me out of the loop.
If you want a mechanical alternative to a hand-trailed ATR band, Keltner channel width based trailing, a Donchian channel trailing stop, or a classic Parabolic SAR all do a similar job. They feel different on the chart, but the principle is identical: scale the stop to volatility, and only ratchet in the direction of the trade.
Common Mistakes With ATR Stop Placement
The most expensive mistake I see is using a multiplier that never changes, regardless of setup. A breakout from a tight base and a late-stage pullback in a noisy trend do not deserve the same distance. Traders who lock in 2.0x ATR for life usually end up with stops that are either too loose on the clean trades (eating back profit) or still too tight on the whippy ones (getting stopped on nothing). The multiplier is part of the trade selection, not a global setting.
The second mistake is using a short ATR lookback. A 5-day or 7-day ATR responds so fast to the latest volatility that it stops being an average at all. On a quiet week, it drops low enough to give you a fake tight stop right before volatility returns. On a loud week, it blows out and forces you to choke the position size right when the setups are cleanest. I have tried 7-day and 10-day variants and always come back to 14 for the same reason Wilder picked it. It is the shortest window that still behaves like an average.
The third mistake is anchoring the band to a round number or an intraday low instead of the entry or latest close. A trader buys AMD at 220, sees that 200 is a nice round level, and drops a stop there because it “feels safe.” That is not a volatility stop. That is a hope stop. If the 200 level is more than 2x ATR away, the position size is wrong. If it is less than 1x ATR away, the stock is going to take you out on any normal down day.
The fourth mistake is ignoring the relationship between the ATR stop and position size. The whole point of volatility-adjusted stops is that they slot into a risk-based position sizing formula. If you are risking a constant dollar amount per trade, the number of shares you buy equals that dollar amount divided by the per-share stop distance. On AMD at 1.5x ATR, that is about 13.38 per share. On TSLA at 2.0x ATR, that is about 29.61 per share. For a 1,000 dollar per trade risk budget, that means 74 AMD shares versus 33 TSLA shares. Both trades risk the same dollars. Neither is the same trade. If you ignore this and buy the same share count in each, the risk is whatever the market decides to hand you, and that is not a system.
The fifth mistake, and the one I make most often, is moving the stop during the day. Recomputing ATR on a 10:30 AM bar, staring at the chart, and deciding that maybe today’s move is “special” is how good rules die. Evening me set the stop for a reason. Daytime me does not get a vote.
Where the ATR Band Earns Its Keep
An ATR stop will not save a bad entry. It will not pick a good breakout. It will not forecast earnings or anticipate a Fed speech. What it does is simple. It makes the cost of being wrong proportional to how much the stock moves on a normal day, which means every loss is roughly the same percentage of your account regardless of which ticker you are trading. That is all any stop rule is supposed to do. The reason 1.5x to 2x ATR wins over fixed percent rules is that it adapts to the stock automatically. You do nothing. The math does the work.
If I could only keep one line on my swing trading charts, it would not be a moving average or an oscillator. It would be a 1.5x ATR band trailing the highest close since entry. Everything else is commentary. The band is the contract I sign with each position, and when the price breaks it, the trade is over. That rule alone has cut more losers short for me than any pattern I ever learned to recognize.
Educational content only. Not investment advice. Trading involves risk. You are responsible for your decisions.